Re: [PATCH 1/3] comm: Introduce comm_lock spinlock to protecttask->comm access

[John Stultz]

On Tue, 2011-05-17 at 23:27 +0200, Ingo Molnar wrote:
> * John Stultz <john.stultz@xxxxxxxxxx> wrote:
> 
> > The implicit rules for current->comm access being safe without locking are no 
> > longer true. Accessing current->comm without holding the task lock may result 
> > in null or incomplete strings (however, access won't run off the end of the 
> > string).
> 
> This is rather unfortunate - task->comm is used in a number of performance 
> critical codepaths such as tracing.
> 
> Why does this matter so much? A NULL string is not a big deal.

I'll defer to KOSAKI Motohiro and David on this bit. :)

> Note, since task->comm is 16 bytes there's the CMPXCHG16B instruction on x86 
> which could be used to update it atomically, should atomicity really be 
> desired.

Could we use this where cmpxchg16b is available and fall back to locking
if not? Or does that put too much of a penalty on arches that don't have
cmpxchg16b support?

Alternatively, we can have locked accessors that are safe in the
majority of slow-path warning printks, and provide unlocked accessors
for cases where the performance is critical and the code can properly
handle possibly incomplete comms.

thanks
-john



--
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@xxxxxxxxxxxxxxx
More majordomo info at  http://vger.kernel.org/majordomo-info.html
Please read the FAQ at  http://www.tux.org/lkml/